Manages data associated with a row/column/block distributed array. More...

#include <systolic.h>

Inheritance diagram for madness::DistributedMatrix< T >:

Collaboration diagram for madness::DistributedMatrix< T >:

Public Member Functions
	DistributedMatrix (World &world, int64_t n, int64_t m, int64_t coltile, int64_t rowtile)

virtual	~DistributedMatrix ()

int64_t	coldim () const
	Returns the column dimension of the matrix ... i.e., n for A(n,m) More...

int64_t	rowdim () const
	Returns the row dimension of the matrix ... i.e., m for A(n,m) More...

int64_t	coltile () const
	Returns the column tile size. More...

int64_t	rowtile () const
	Returns the row tile size. More...

int64_t	process_coldim () const
	Returns the no. of processors in the column dimension. More...

int64_t	process_rowdim () const
	Returns the no. of processors in the row dimension. More...

int64_t	local_size () const
	Returns the total no. of elements stored on this processor. More...

int64_t	local_coldim () const
	Returns the no. of column elements stored on this processor. More...

int64_t	local_rowdim () const
	Returns the no. of row elements stored on this processor. More...

void	local_colrange (int64_t &ilow, int64_t &ihigh) const
	Returns the inclusive range of column indices on this processor. More...

void	local_rowrange (int64_t &jlow, int64_t &jhigh) const
	Returns the inclusive range of row indices on this processor. More...

void	get_colrange (int p, int64_t &ilow, int64_t &ihigh) const
	Returns the inclusive range of column indices on processor p. More...

World &	get_world () const
	Returns associated world. More...

Tensor< T > &	data ()
	Returns reference to data. More...

const Tensor< T > &	data () const
	Returns const reference to data. More...

bool	is_column_distributed () const
	Returns true if the matrix is column distributed (i.e., row dimension not distributed) More...

bool	is_row_distributed () const
	Returns true if the matrix is row distributed (i.e., column dimension not distributed) More...

void	copyin (const Tensor< T > &s)
	Given the full matrix(n,m), copy in the data range local to this processor. More...

void	copyout (Tensor< T > &s) const
	Given the full matrix s(n,m), copy out the data range local to this processor. More...

	DistributedMatrix ()
	Default constructor makes an empty matrix that cannot be used except as a target for assignemnt. More...

	DistributedMatrix (const DistributedMatrixDistribution &d)
	Constructs a distributed matrix with given distribution info. More...

	DistributedMatrix (const DistributedMatrix< T > &A, bool deepcopy=false)
	Copy constructor copies dimensions, distribution, and shallow copy of content (unless deepcopy=true) More...

DistributedMatrix< T > &	operator= (const DistributedMatrix< T > &A)
	Assigment copies dimensions, distribution, and shallow copy of content. More...

virtual	~DistributedMatrix ()

void	clear ()
	Frees memory and resets state to same as default constructor. More...

template<typename funcT >
void	fill (const funcT &f)
	Fills the matrix with the provided function of the indices. More...

void	fill (T value)
	Fills the matrix with a scalar. More...

void	fill_identity ()

Tensor< T > &	data ()
	Returns reference to the local data. More...

const Tensor< T > &	data () const
	Returns const reference to data. More...

void	copy_from_replicated (const Tensor< T > &s)
	Copy from the replicated (m,n) matrix into the distributed matrix. More...

void	copy_to_replicated (Tensor< T > &s) const
	Copy from the distributed (m,n) matrix into the replicated matrix (collective call) More...

void	copy_from_replicated_patch (int64_t ilow, int64_t ihigh, int64_t jlow, int64_t jhigh, const Tensor< T > &s)
	Copy from replicated patch (inclusive index range) into the distributed matrix. More...

void	copy_to_replicated_patch (int64_t ilow, int64_t ihigh, int64_t jlow, int64_t jhigh, Tensor< T > &s) const
	Copy from distributed matrix into replicated patch (inclusive index range; collective call) More...

void	extract_columns (int64_t jlow, int64_t jhigh, DistributedMatrix< T > &U) const

template<typename R >
bool	has_same_dimension_and_distribution (const DistributedMatrix< R > &A)

DistributedMatrix< T > &	operator+= (const DistributedMatrix< T > &A)
	Inplace addition — dimensions and distribution must be identical. More...

DistributedMatrix< T >	operator+ (const DistributedMatrix< T > &A) const
	Out of place addition — dimensions and distribution must be identical. More...

DistributedMatrix< T > &	operator*= (const T s)
	Inplace scale by a constant. More...

void	set (int64_t i, int64_t j, const T x)
	Sets element (i,j) to v if (i,j) is local, otherwise throws MadnessException. More...

T	get (int64_t i, int64_t j) const
	Gets element (i,j) if (i,j) is local, otherwise throws MadnessException. More...

Public Member Functions inherited from madness::DistributedMatrixDistribution
	DistributedMatrixDistribution ()
	Default constructor makes an invalid distribution. More...

void	clear ()
	Resets state to same as default constructor. More...

bool	operator== (const DistributedMatrixDistribution &d) const

int64_t	coldim () const
	Returns the column dimension of the matrix ... i.e., n for A(n,m) More...

int64_t	rowdim () const
	Returns the row dimension of the matrix ... i.e., m for A(n,m) More...

int64_t	coltile () const
	Returns the column tile size. More...

int64_t	rowtile () const
	Returns the row tile size. More...

int64_t	process_coldim () const
	Returns the no. of processors in the column dimension. More...

int64_t	process_rowdim () const
	Returns the no. of processors in the row dimension. More...

int64_t	local_size () const
	Returns the total no. of elements stored on this processor. More...

int64_t	local_coldim () const
	Returns the no. of column elements stored on this processor. More...

int64_t	local_rowdim () const
	Returns the no. of row elements stored on this processor. More...

void	local_colrange (int64_t &ilow, int64_t &ihigh) const
	Returns the inclusive range of column indices on this processor. More...

void	local_rowrange (int64_t &jlow, int64_t &jhigh) const
	Returns the inclusive range of row indices on this processor. More...

int64_t	local_ilow () const
	Returns the first column index on this processor (0 if no data present) More...

int64_t	local_ihigh () const
	Returns the last column index on this processor (-1 if no data present) More...

int64_t	local_jlow () const
	Returns the first row index on this processor (0 if no data present) More...

int64_t	local_jhigh () const
	Returns the last row index on this processor (0 if no data present) More...

void	get_range (int p, int64_t &ilow, int64_t &ihigh, int64_t &jlow, int64_t &jhigh) const
	Returns the inclusive ranges of column and row indicies on processor p. More...

void	get_colrange (int p, int64_t &ilow, int64_t &ihigh) const
	Returns the inclusive range of column indices on processor p. More...

void	get_rowrange (int p, int64_t &jlow, int64_t &jhigh) const
	Returns the inclusive range of row indices on processor p. More...

World &	get_world () const
	Returns the associated world. More...

bool	is_column_distributed () const
	Returns true if the matrix is column distributed (i.e., row dimension not distributed) More...

bool	is_row_distributed () const
	Returns true if the matrix is row distributed (i.e., column dimension not distributed) More...

const DistributedMatrixDistribution &	distribution () const
	Returns the distribution (aka *this) More...

ProcessID	owner (int64_t i, int64_t j) const
	Returns the number of the process that owns element (i,j) More...

virtual	~DistributedMatrixDistribution ()

Protected Member Functions
	DistributedMatrix (World &world, int64_t n, int64_t m, int64_t coltile, int64_t rowtile)
	Constructs a distributed matrix dimension (n,m) with specified tile sizes and initialized to zero. More...

Protected Member Functions inherited from madness::DistributedMatrixDistribution
	DistributedMatrixDistribution (World &world, int64_t n, int64_t m, int64_t coltile, int64_t rowtile)
	Constructs distribution and size info for a matrix (for use by factory functions only) More...

Friends
DistributedMatrix< T >	interleave_rows (const DistributedMatrix< T > &a, const DistributedMatrix< T > &b)
	Generates a distributed matrix with rows of `a` and `b` interleaved. More...

DistributedMatrix< T >	concatenate_rows (const DistributedMatrix< T > &a, const DistributedMatrix< T > &b)
	Generates a distributed matrix with rows of `a` and `b` concatenated. More...

Additional Inherited Members
Protected Attributes inherited from madness::DistributedMatrixDistribution
World *	pworld

int64_t	P

ProcessID	rank

int64_t	n

int64_t	m

int64_t	tilen

int64_t	tilem

int64_t	Pcoldim

int64_t	Prowdim

int64_t	Pcol

int64_t	Prow

int64_t	ilo

int64_t	ihi

int64_t	jlo

int64_t	jhi

int64_t	idim

int64_t	jdim

Detailed Description

template<typename T>
class madness::DistributedMatrix< T >

Manages data associated with a row/column/block distributed array.

The class itself provides limited functionality for accessing the data and is primarily intended to provide base functionality for use by matrix algorithms and other matrix classes.

The constructor is deliberately simple. Factory functions are expected to be the main construction tool.

The class itself provides limited functionality for accessing the data and is primarily intended to provide base functionality for use by matrix algorithms and other matrix classes.

The constructor is deliberately simple. Factory functions are expected to be the main construction tool.

Assignment and copy are shallow just like for tensor and for the same reasons.

To get a deep copy use the copy function (again just like for tensors).

Constructor & Destructor Documentation

template<typename T>

madness::DistributedMatrix< T >::DistributedMatrix	(	World &	world,
		int64_t	n,
		int64_t	m,
		int64_t	coltile,
		int64_t	rowtile
	)

inline

template<typename T>

virtual madness::DistributedMatrix< T >::~DistributedMatrix ( )

inlinevirtual

template<typename T>

madness::DistributedMatrix< T >::DistributedMatrix	(	World &	world,
		int64_t	n,
		int64_t	m,
		int64_t	coltile,
		int64_t	rowtile
	)

inlineprotected

Constructs a distributed matrix dimension (n,m) with specified tile sizes and initialized to zero.

[deprecated ... use factory functions instead]

The matrix is tiled over a grid of processes as specified by the tile sizes.

Parameters

[in]	World	The world
[in]	n	The matrix column dimension
[in]	m	The matrix row dimension
[in]	coltile	Tile size for the columns
[in]	rowtile	Tile size for the rows

template<typename T>

madness::DistributedMatrix< T >::DistributedMatrix ( )

inline

Default constructor makes an empty matrix that cannot be used except as a target for assignemnt.

template<typename T>

madness::DistributedMatrix< T >::DistributedMatrix ( const DistributedMatrixDistribution & d )

inline

Constructs a distributed matrix with given distribution info.

template<typename T>

madness::DistributedMatrix< T >::DistributedMatrix	(	const DistributedMatrix< T > &	A,
		bool	deepcopy = `false`
	)

inline

Copy constructor copies dimensions, distribution, and shallow copy of content (unless deepcopy=true)

template<typename T>

virtual madness::DistributedMatrix< T >::~DistributedMatrix ( )

inlinevirtual

Member Function Documentation

template<typename T>

void madness::DistributedMatrix< T >::clear ( )

inline

Frees memory and resets state to same as default constructor.

template<typename T>

int64_t madness::DistributedMatrix< T >::coldim ( ) const

inline

Returns the column dimension of the matrix ... i.e., n for A(n,m)

Referenced by madness::concatenate_columns(), madness::concatenate_rows(), madness::DistributedMatrix< double >::extract_columns(), madness::idMatrix(), madness::interleave_rows(), matrix_inner(), madness::matrix_inner(), and madness::transform().

template<typename T>

int64_t madness::DistributedMatrix< T >::coltile ( ) const

inline

Returns the column tile size.

Referenced by madness::concatenate_rows(), madness::DistributedMatrix< double >::extract_columns(), madness::idMatrix(), madness::interleave_rows(), and madness::SystolicMatrixAlgorithm< double >::SystolicMatrixAlgorithm().

template<typename T>

void madness::DistributedMatrix< T >::copy_from_replicated ( const Tensor< T > & s )

inline

Copy from the replicated (m,n) matrix into the distributed matrix.

Parameters

[in] The input tensor

template<typename T>

void madness::DistributedMatrix< T >::copy_from_replicated_patch	(	int64_t	ilow,
		int64_t	ihigh,
		int64_t	jlow,
		int64_t	jhigh,
		const Tensor< T > &	s
	)

inline

Copy from replicated patch (inclusive index range) into the distributed matrix.

Parameters

[in]	ilow	First `i` index in patch
[in]	ihigh	Last `i` index in patch
[in]	jlow	First `j` index in patch
[in]	jhjgh	Last `j` index in patch

Referenced by madness::matrix_inner(), and matrix_inner().

template<typename T>

void madness::DistributedMatrix< T >::copy_to_replicated ( Tensor< T > & s ) const

inline

Copy from the distributed (m,n) matrix into the replicated matrix (collective call)

The entire output tensor is zeroed, the local data copied into it, and then a global sum performed to replicate the data.

Parameters

[out] The output tensor (assumed already allocated to be at least large enough in both dimensions)

Referenced by distributed_localize_PM(), madness::SCF::initial_guess(), madness::SCF::make_fock_matrix(), and madness::transform().

template<typename T>

void madness::DistributedMatrix< T >::copy_to_replicated_patch	(	int64_t	ilow,
		int64_t	ihigh,
		int64_t	jlow,
		int64_t	jhigh,
		Tensor< T > &	s
	)		const

inline

Copy from distributed matrix into replicated patch (inclusive index range; collective call)

The entire output tensor is zeroed, relevant local data copied into it, and then a global sum performed to replicate the data.

Parameters

[in]	ilow	First `i` index in patch
[in]	ihigh	Last `i` index in patch
[in]	jlow	First `j` index in patch
[in]	jhjgh	Last `j` index in patch

template<typename T>

void madness::DistributedMatrix< T >::copyin ( const Tensor< T > & s )

inline

Given the full matrix(n,m), copy in the data range local to this processor.

Referenced by main(), and plocalize_boys().

template<typename T>

void madness::DistributedMatrix< T >::copyout ( Tensor< T > & s ) const

inline

Given the full matrix s(n,m), copy out the data range local to this processor.

Referenced by main().

template<typename T>

Tensor<T>& madness::DistributedMatrix< T >::data ( )

inline

Returns reference to data.

Referenced by madness::concatenate_columns(), madness::concatenate_rows(), madness::interleave_rows(), madness::SCF::solve(), and madness::SystolicMatrixAlgorithm< double >::SystolicMatrixAlgorithm().

template<typename T>

const Tensor<T>& madness::DistributedMatrix< T >::data ( ) const

inline

Returns const reference to data.

template<typename T>

Tensor<T>& madness::DistributedMatrix< T >::data ( )

inline

Returns reference to the local data.

The local data is a tensor dimension (local_coldim,local_rowdim) and if either of the dimensions are zero there is no data. A natural way to loop thru the data that gives you the actual row and column indices is

const Tensor<T>& t = A.data();
for (int64_t i=A.get_ilow(); i<=A.get_ihigh(); i++) {
   for (int64_t j=A.get_jlow(); j<=A.get_jhigh(); j++) {
       the ij'th element is t(i-ilo, j-jlo)
   }
}

Returns: Reference to non-constant tensor holding the local data

template<typename T>

const Tensor<T>& madness::DistributedMatrix< T >::data ( ) const

inline

Returns const reference to data.

The local data is a tensor dimension (local_coldim,local_rowdim) and if either of the dimensions are zero there is no data. A natural way to loop thru the data that gives you the actual row and column indices is

const Tensor<T>& t = A.data();
for (int64_t i=A.get_ilow(); i<=A.get_ihigh(); i++) {
   for (int64_t j=A.get_jlow(); j<=A.get_jhigh(); j++) {
       the ij'th element is t(i-ilo, j-jlo)
   }
}

Returns: Reference to constant tensor holding the local data

template<typename T>

void madness::DistributedMatrix< T >::extract_columns	(	int64_t	jlow,
		int64_t	jhigh,
		DistributedMatrix< T > &	U
	)		const

inline

Referenced by distributed_localize_PM().

template<typename T>

template<typename funcT >

void madness::DistributedMatrix< T >::fill ( const funcT & f )

inline

Fills the matrix with the provided function of the indices.

Parameters

[in] f The matrix is filled using a[i,j]=f(i,j)

Referenced by madness::DistributedMatrix< double >::fill_identity().

template<typename T>

void madness::DistributedMatrix< T >::fill ( T value )

inline

Fills the matrix with a scalar.

Parameters

[in] value The matrix is filled using a[i,j]=value

template<typename T>

void madness::DistributedMatrix< T >::fill_identity ( )

inline

Referenced by distributed_localize_PM().

template<typename T>

T madness::DistributedMatrix< T >::get	(	int64_t	i,
		int64_t	j
	)		const

inline

Gets element (i,j) if (i,j) is local, otherwise throws MadnessException.

template<typename T>

void madness::DistributedMatrix< T >::get_colrange	(	int	p,
		int64_t &	ilow,
		int64_t &	ihigh
	)		const

inline

Returns the inclusive range of column indices on processor p.

If there is no data on this processor it returns ilow=0 and ihigh=-1

Referenced by madness::SystolicMatrixAlgorithm< double >::SystolicMatrixAlgorithm().

template<typename T>

World& madness::DistributedMatrix< T >::get_world ( ) const

inline

Returns associated world.

Referenced by madness::concatenate_columns(), madness::concatenate_rows(), madness::DistributedMatrix< double >::copy_to_replicated(), madness::DistributedMatrix< double >::copy_to_replicated_patch(), madness::SystolicMatrixAlgorithm< double >::get_world(), madness::idMatrix(), madness::interleave_rows(), matrix_inner(), madness::matrix_inner(), madness::SystolicMatrixAlgorithm< double >::solve(), and madness::SystolicMatrixAlgorithm< double >::solve_sequential().

template<typename T>

template<typename R >

bool madness::DistributedMatrix< T >::has_same_dimension_and_distribution ( const DistributedMatrix< R > & A )

inline

Referenced by madness::DistributedMatrix< double >::operator+(), and madness::DistributedMatrix< double >::operator+=().

template<typename T>

bool madness::DistributedMatrix< T >::is_column_distributed ( ) const

inline

Returns true if the matrix is column distributed (i.e., row dimension not distributed)

Referenced by madness::concatenate_rows(), madness::DistributedMatrix< double >::extract_columns(), and madness::SystolicMatrixAlgorithm< double >::SystolicMatrixAlgorithm().

template<typename T>

bool madness::DistributedMatrix< T >::is_row_distributed ( ) const

inline

Returns true if the matrix is row distributed (i.e., column dimension not distributed)

Referenced by madness::concatenate_columns().

template<typename T>

int64_t madness::DistributedMatrix< T >::local_coldim ( ) const

inline

Returns the no. of column elements stored on this processor.

template<typename T>

void madness::DistributedMatrix< T >::local_colrange	(	int64_t &	ilow,
		int64_t &	ihigh
	)		const

inline

Returns the inclusive range of column indices on this processor.

If there is no data on this processor it returns ilow=0 and ihigh=-1

Referenced by madness::idMatrix().

template<typename T>

int64_t madness::DistributedMatrix< T >::local_rowdim ( ) const

inline

Returns the no. of row elements stored on this processor.

template<typename T>

void madness::DistributedMatrix< T >::local_rowrange	(	int64_t &	jlow,
		int64_t &	jhigh
	)		const

inline

Returns the inclusive range of row indices on this processor.

If there is no data on this processor it returns jlow=0 and jhigh=-1

template<typename T>

int64_t madness::DistributedMatrix< T >::local_size ( ) const

inline

Returns the total no. of elements stored on this processor.

Referenced by madness::concatenate_columns(), madness::concatenate_rows(), madness::DistributedMatrix< double >::copy_from_replicated(), madness::DistributedMatrix< double >::copy_to_replicated(), madness::DistributedMatrix< double >::copyin(), and madness::DistributedMatrix< double >::copyout().

template<typename T>

DistributedMatrix<T>& madness::DistributedMatrix< T >::operator*= ( const T s )

inline

Inplace scale by a constant.

Parameters

[in] s The scaling factor

Returns: A reference to the current matrix

template<typename T>

DistributedMatrix<T> madness::DistributedMatrix< T >::operator+ ( const DistributedMatrix< T > & A ) const

inline

Out of place addition — dimensions and distribution must be identical.

Parameters

[in] A The matrix to add to the current matrix

Returns: A new matrix with the same dimensions and distribution as the inputs

template<typename T>

DistributedMatrix<T>& madness::DistributedMatrix< T >::operator+= ( const DistributedMatrix< T > & A )

inline

Inplace addition — dimensions and distribution must be identical.

Parameters

[in] A The matrix to add to the current matrix

Returns: A reference to the current matrix

template<typename T>

DistributedMatrix<T>& madness::DistributedMatrix< T >::operator= ( const DistributedMatrix< T > & A )

inline

Assigment copies dimensions, distribution, and shallow copy of content.

template<typename T>

int64_t madness::DistributedMatrix< T >::process_coldim ( ) const

inline

Returns the no. of processors in the column dimension.

Referenced by madness::DistributedMatrix< double >::get_colrange(), and madness::DistributedMatrix< double >::is_row_distributed().

template<typename T>

int64_t madness::DistributedMatrix< T >::process_rowdim ( ) const

inline

Returns the no. of processors in the row dimension.

Referenced by madness::DistributedMatrix< double >::get_colrange(), and madness::DistributedMatrix< double >::is_column_distributed().

template<typename T>

int64_t madness::DistributedMatrix< T >::rowdim ( ) const

inline

Returns the row dimension of the matrix ... i.e., m for A(n,m)

Referenced by madness::concatenate_columns(), madness::concatenate_rows(), madness::DistributedMatrix< double >::extract_columns(), madness::idMatrix(), madness::interleave_rows(), matrix_inner(), madness::matrix_inner(), and madness::transform().

template<typename T>

int64_t madness::DistributedMatrix< T >::rowtile ( ) const

inline

Returns the row tile size.

Referenced by madness::concatenate_columns(), madness::idMatrix(), and madness::interleave_rows().

template<typename T>

void madness::DistributedMatrix< T >::set	(	int64_t	i,
		int64_t	j,
		const T	x
	)

inline

Sets element (i,j) to v if (i,j) is local, otherwise throws MadnessException.

Friends And Related Function Documentation

template<typename T>

DistributedMatrix<T> concatenate_rows	(	const DistributedMatrix< T > &	a,
		const DistributedMatrix< T > &	b
	)

friend

Generates a distributed matrix with rows of a and b concatenated.

Generates a column-distributed matrix with rows of a and b contatenated.

I.e., c[i,j] = a[i,j] if n<na or b[i,j] if j>=na

The matrices a and b must have the same column size (i.e., the same number of rows) and be column distributed with the same column tilesze. The result is also column distributed with the same column tilesize as the input matrices.

I.e., c[i,j] = a[i,j] if j<na or b[i,j-na] if j>=na

The matrices a and b must have the same column size (i.e., the same number of rows) and be column distributed with the same column tilesze. The result is also column distributed with the same column tilesize as the input matrices.

Parameters

[in]	a	The first matrix
[in]	b	The second matrix

Returns: The result matrix

template<typename T>

DistributedMatrix<T> interleave_rows	(	const DistributedMatrix< T > &	a,
		const DistributedMatrix< T > &	b
	)

friend

Generates a distributed matrix with rows of a and b interleaved.

I.e., the even rows of the result will be rows of a , and the odd rows those of b .

The matrices a and b must have the same dimensions and be identically distributed. The result will have a doubled column dimension and column tile size. The row dimension is unchanged.

I.e., the even rows of the result will be rows of a , and the odd rows those of b .

The matrices a and b must have the same dimensions and be identically distributed. The result will have a doubled column dimension and column tile size. The row dimension is unchanged.

Parameters

[in]	a	The matrix providing even rows of the result
[in]	b	The matrix providing odd rows of the result

Returns: The result matrix

The documentation for this class was generated from the following files:

Public Member Functions

Protected Member Functions

Friends

Additional Inherited Members

Detailed Description

template<typename T> class madness::DistributedMatrix< T >

Constructor & Destructor Documentation

Member Function Documentation

Friends And Related Function Documentation

template<typename T>
class madness::DistributedMatrix< T >